A Landslide Susceptibility Mapping Method Based on Geographic Information System and Data Enhancement Techniques: A Case Study of Guangzhou City, China

Landslides are one of the most widespread and hazardous geologic hazards in the world. Landslide susceptibility mapping (LSM) is an effective way to identify landslide-prone areas to prevent and reduce landslide hazards. However, the accuracy of LSM is greatly limited by the balance of landslide data. The collection of high-quality landslide inventories is labor-intensive. In this paper, four data enhancement techniques are used to correct the unbalanced landslide dataset based on GIS. The method is applied to LSM in Guangzhou City. And the difference in the accuracy of the enhancement is compared with two assessment models, SVM and RF. The experimental results show that the data enhancement technique helps to improve the accuracy of the evaluated models, and the evaluation results of all models show the best performance by GAN-RF. This study shows that the balance of landslide data greatly affects the accuracy of the assessment model. And the data enhancement technique enhances the robustness of the assessment model and improves the accuracy of the prediction results.